To start, I would first like to ask you a few questions.
How many of you have heard of the Digital Public Library of America before?
How many of you have used DPLA?
If you're not familiar with DPLA, that's alright - I do understand that I'm in Canada. The Digital Public Library of America brings together the riches of America’s libraries, archives, and museums, and makes them freely available to the world.
It strives to contain the full breadth of human expression, from the written word, to works of art and culture, to records of America’s heritage, to the efforts and data of science.
The "full range of human expression" means all kinds of stuff—cultural heritage materials such as images, sound files, moving images, journals, books, data sets, etc.
The DPLA aims to expand this crucial realm of openly available materials, and make those riches more easily discovered and more widely usable and used.
DPLA has a unique opportunity to connect people across the United States and its territories to connect with cultural heritage materials. In this presentation, I will be talking about some ways in which DPLA has provided a local connection or outreach, and ways in which we could improve that going forward.

DPLA didn’t just appear one day as a small start-up non-profit.
Instead, it was the result of a two-year grassroots planning initiative that started in 2010.
The DPLA planning process began in October 2010 at a meeting in Cambridge, MA.
During this meeting, 40 leaders from libraries, foundations, academia, and technology projects agreed to work together to create "an open, distributed network of comprehensive online resources that would draw on the nation’s living heritage from libraries, universities, archives, and museums in order to educate, inform, and empower everyone in current and future ­generations."
That single sentence banded together hundreds toward a common goal of building a national digital library platform.

In December 2010, the Berkman Center for Internet & Society at Harvard University convened leading experts in libraries, technology, law, and education to begin work on this ambitious project.
A two-year process of intense grassroots community organization, beginning in October 2011 and hosted at the Berkman Center under the aegis of the DPLA Secretariat, brought together hundreds of public and research librarians, innovators, digital humanists, and other volunteers—organized into six workstreams and led by a distinguished Steering Committee—helped to scope, design, and construct the DPLA. The culmination of all of this hard work was DPLA’s successful launch on April 18, 2013. While the actual in-person celebration event had to be canceled on account of the Boston Marathon bombing, DPLA’s site and services launched right on-time.
The April launch also marked the transition from the Harvard-based planning phase to what we call DPLA’s operational phase, or the start of DPLA as an independent 501c3 non-profit organization.

The DPLA platform is one of the most important parts of our technical infrastructure.
It provides us as well as our users with the ability to search and retrieve metadata ingested from our service hubs and content hubs.
In fact, the Platform directly provides this functionality to the DPLA portal.
Most importantly, we provide free and open access to the Platform and the metadata available from within it.

For most of American history, the ability to access materials for free through public libraries has been a central part of our culture.
The DPLA works with others to ensure that this critical, open intellectual landscape remains vibrant and broad in the face of increasingly restrictive digital options.
The DPLA seeks to multiply openly accessible materials to strengthen the public option that libraries represent in their communities.

As an aggregator of metadata from many institutions, DPLA is in a unique position to help our partners recognize and manage data quality issues.
In October 2013, Europeana and the DPLA organized a first joint rights management workshop to explore this possibility in Boston, Massachusetts.
As a result of further discussions between the DPLA and Europeana, a small joint working group to explore the possibilities for concrete collaborations in more detail has been established. This important work will, above all, make rights clear to the end user and provide a framework for aggregators and our partners.
With the creation and standardization of actionable rights statements, users will know when a work is in the Public Domain, covered under a Creative Commons license or is Rights Restricted, among other possible labels. In addition, we recently submitted an entry to the Knight Foundation's Knight News Challenge to potentially support this work.

Now, you may wonder how this is possible, because those 5.8 million items had to come from somewhere.

We have a partnership model which we refer to as our "hubs model."
The Hubs Program is designed to establish a national network of state and regional digital libraries, as well as large institutional digital libraries.
The hubs model allows us to bring together digitized content from across the country into a single access point for end users, and an open platform for developers.

As you can see, the hubs model allows local collections to become more easily discoverable.
The hubs that assist us with bringing that content, as well as the partners that provide their hub with content, both serve a vital function.
The hubs help mediate the relationship between the partners and DPLA.
The partners are the local institutions with direct expertise and knowledge in the collections and the related subjects.
Specifically, the partners and hubs provide an important local connection.

Of course, the hubs model allows us to do one kind of outreach.
In addition to undertaking outreach to librarians and archivists, DPLA should also provide outreach to the general public However, the DPLA staff is small, with seven current employees. How can we improve outreach to new communities with this limitation, plus further limitations on our travel budgets?

In September 2013, DPLA staff conceived of a program of community representatives, or volunteers that would help spread the word about DPLA.
While DPLA was still in its planning phase, my colleagues found that our web forums and committees provided a forum for interested and motivated people to help out and give us feedback.

The Community Reps program is already showing some early signs of success, and we could not keep up with the enthusiasm shown by the community!
We have just announced recruitment for our second class of community reps, and we are looking to greatly extend our reach by searching for applicants from states and territories in which we have no reps.
Applications close on April 30th, and we look forward to further expanding our reach and geographic diversity.
Given some inital review of the Community Reps program, we have decided to more explicitly ask applicants to identify the communities to which they intend to provide outreach.
Doing so will allow us to have a better understanding of where our strengths and gaps may be, and can provide important information about where to undertake targeted outreach in the future, either through the Reps program or with DPLA staff.

Finally, and perhaps the most obvious way in which DPLA provides a place based location is through our map. The map interface that allows users to see records organized by geolocation.
It is important to note that not all records in DPLA have geolocation information.
As the search box at the top indicates, only records with geolocation info appear on the map, and while that is a large portion of the DPLA’s collection, it is not all items related to a search.

One very common misconception about the DPLA map is that it organizes records geographically by the institutions they come from.
For example, users of this search might assumed that we have 56 "baseball" items owned by Arizona institutions.
In reality, we have 56 items related to baseball that have been geolocated to Arizona – like images taken there.
These records do not necessarily come from Arizona partners.

As you zoom in, you can more clearly see the places depicted or represented in these items, based on the information.
But how do we get this information?

This goes back to the process by which we work with the metadata we receive.
Many of our providers include place names or other geographic headings in their metadata, and as part of the enrichment process we identify those headings.
We send the text of those headings to a geocoding service, which looks them up in a database, and returns a set of latitude and longitude coordinates for that place.
Once we have those coordinates, we send them back to another service that "reverse geocodes" them and provides us with a full place hierarchy, including country, state or province, region or county, and city, when available.
Usually, this works well, except when it doesn't.

One example of this is what I call the Kansas problem.
If you look at the map, it suggests that there are nearly 500,000 records that are associated with Kansas.
However, very few (if any) of these items are about Kansas. So, why is this happening? Naturally, we discovered it was a bug in our process. The only spatial value being passed to us from the Hubs is "United States."
When we get ONLY that value, we're grabbing coordinates for the center of the country and then reverse identifying the location... as Kansas. We've had other problems as well, such as geocoding services misidentifying which place we were talking about.

As a small, non-profit organization, we may not always have the subject or regional expertise about a topic to know whether the information is accurate.
However, as a national-level aggregator, DPLA has some degree of responsibility here. But what is that level of responsibility? We do have some degree of influence, but we cannot always control what our hubs and partners do.
How should we be gathering feedback from users about the accuracy of our data?
In addition, how should that feedback flow back to the hubs and the partners from which we received the data?
We do not yet have the solutions, but we're certainly looking for suggestions.